Granulated bentonite mixtures (GBMs) have been regarded as effective buffer materials in the deep geological disposal of radioactive waste due to their operational advantages, such as ease of transportation and in-situ placement/backfilling. Many studies have been done to characterize the hydraulic and thermal properties of GBMs as well as their swelling properties. Only limited studies, however, have investigated their gas transport properties, even though these properties affect their compactness during in-situ placement/backfilling and subsequent gas diffusion and advection in the buffer zone. The aim of this study is to understand the gas transport parameters, i.e., air permeability (ka) and gas diffusivity (Dp/Do), of tested samples compacted at different dry densities (DDs) under air-dried conditions, linking them with the measured density distribution characteristics determined by microfocus X-ray computed tomography (MFXCT) analysis. Two types of GBMs were used in this study: 1) FE-GBM (prepared from National Standard® bentonite, Wyoming, USA): this material was used in the Full-scale Emplacement (FE) experiment at the Mont Terri rock laboratory, Switzerland) and 2) OK-GBM (prepared from a bentonite, originating from Japan, with the trade name of OK bentonite, Kunimine Industries). The tested samples were firstly packed in a 100-cm3 acrylic core with different DDs, ranging from loose to dense (1.09 to 1.75 g/cm3), and scanned by MFXCT. The weighting factors, wf (fine fraction; lower density) and wc (coarse fraction; higher density) (wf + wc = 1), were determined after the peak separation of the measured CT brightness histograms from the reconstructed three-dimensional multiplanar reconstruction (MPR) images of the MFXCT analysis. The measured ka and Dp/Do were highly dependent on the DDs, the ka (ɛ) values fitted well with a power law model, and the Dp/Do (ɛ) was predicted accurately by several previously proposed models. For both FE-GBM and OK-GBM, there were good linear relationships between the gas transport parameters and wc × DD, implying that the weight of the coarse fraction controlled ka and Dp/Do. Moreover, the Kozeny-Carman model, incorporating the measured volumetric surfaces from the MFXCT analysis, was able to predict the ka values well.
Read full abstract